1. No category

Agricultural Systems 137 (2015) 24–38
Contents lists available at ScienceDirect
Agricultural Systems
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / a g s y
Production potential of Lentil (Lens culinaris Medik.) in East Africa
Michel Edmond Ghanem a,1, Hélène Marrou b,c,1, Chandrashekhar Biradar a,
Thomas R. Sinclair b,*
a
b
c
International Center for Agricultural Research in the Dry Areas (ICARDA), North-Africa Platform, Rabat, Morocco
Department of Crop Science, North Carolina State University (NCSU), Raleigh, North Carolina, USA
UMR System, Montpellier SupAgro, Montpellier, France
A R T I C L E
I N F O
Article history:
Received 1 September 2014
Received in revised form 1 December 2014
Accepted 13 March 2015
Available online
Keywords:
East Africa
Lentil
Model
Yield
Sowing date
Phenology
A B S T R A C T
Lentil (Lens culinaris Medik.) could possibly become a major crop in East Africa due to its many uses as
a food and feed. Also, its ability to undertake symbiotic nitrogen fixation is an advantage over cereal crops.
This study simulated lentil yield potential in order to determine the geographical areas in East Africa
that offer potential for consistent lentil production. Results show that there is potential to further expand
the geographical area in which lentil is currently grown in East Africa into Uganda, Kenya, Tanzania and
even Somalia. Response to a change in management practices on potential yield of lentil as a result of
different sowing dates was also examined. In addition, the effect of phenology on yield potential was
examined by comparing a short-season type vs. a long-season type. Delaying sowing alone or in combination with a long-season genotype can result in a high probability of crop yield increase in East Africa.
For the long-season genotype, an optimum sowing window was found between June and July (152–229
day of year) for areas to the north of the Rift Valley. Later sowing dates (229–243 day of year) were found
to be optimal in southern areas of East Africa. These simulations indicated that selection and breeding
for lentil accessions in East Africa should consider changes in plant phenology and/or sowing dates.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
East African countries are dealing with myriads of socioeconomic and environmental problems including poverty, alarming
population growth, low agricultural productivity, land resource degradation, frequent drought and high-heat episodes, leading to limited
crop production. In the last decades, droughts caused widespread
famine and economic hardships in many countries of the region.
Future climate change may lead to altered frequency or severity of
such extreme weather events, potentially worsening water-deficit
events.
Cool-season grain legumes are important protein-rich food crops
of East Africa. Lentil (Lens culinaris Medik.), faba bean (Vicia faba
L.), and chickpea (Cicer arietinum L.) provide a considerable portion
of the diet of the people in this region. For example, in Rwanda bean
supplies 65% of the national dietary protein, compared to 4% from
animal protein, and 32% of the energy (Kelly, 2004). In addition, these
crops can play a major role in sustaining soil fertility by symbiotic
nitrogen fixation because in this region there are major economic
limitations in the use of commercial nitrogen fertilizers. Enhancing and diversifying grain legume production in East Africa would
* Corresponding author. Tel.: 919-513-1620; fax: 919-515-3332.
E-mail address: [email protected] (T.R. Sinclair).
1 H. Marrou and M.E. Ghanem contributed equally to this work.
http://dx.doi.org/10.1016/j.agsy.2015.03.005
0308-521X/© 2015 Elsevier Ltd. All rights reserved.
help in alleviating problems of malnutrition and may bring future
stability to the region (Siddiq and Uebersax, 2012; Solh, 1996).
Lentil is an annual food legume highly valued for grain in the world.
The crop has great significance in cereal-based cropping systems
because of its nitrogen fixing ability, the high protein and high micronutrients seeds for human diet, and its straw for animal feed.
Major producing regions of lentil are in South and West Asia,
Northern Africa, Canada, Australia and the USA (Chen et al., 2011).
Within East Africa, cowpea (Vigna unguiculata L.), groundnut (Arachis
hypogaea L.), common bean (Phaseolus vulgaris L.), soybean (Glycine
max L.), faba bean, and pigeon pea (Cajanus cajan L.) are the most
important grain legumes crops (Timko et al., 2007). Lentil is now
mainly grown in Ethiopia, Sudan, South-Sudan, and Eritrea, but is
not a crop in Kenya, Uganda, Somalia, and Tanzania (Bejiga and
Degago, 2000). Despite this fact, wild relatives of lentil such as
L. ervoides have been reported on the slopes of mount Muhavura,
in the southwest corner of Uganda (Kisoro region) south of the
Equator (Ladzinsky and Smartt, 2000). East African production of
lentil is characterized by a low mean yield of 0.1 t·ha−1compared to
that of Central Asia and North Africa (0.6 t·ha−1) and South and Southeast Asia (1.7 t·ha−1) (FAOSTAT, 2011). The major reason for these
low yields appears to be a result of production on marginal lands
in arid and semi-arid environments without irrigation, weeding or
pest control (Bejiga and Degago, 2000).
There may be, however, an opportunity of growing cool-season
grain legumes like lentil in countries south of Ethiopia such as
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
25
Tanzania by replacing the fallow at the end of the rainy season. In
this case, lentil would not compete with other crops like maize
(Zea mays L.), which are grown during the rainy season (Pundir et al.,
1996). No reports were found exploring this possibility. This paper
explores the possibility for expanding lentil cultivation in East Africa,
toward the southern areas (latitude 22° N to 11°50′ S and longitude 21°50′ E to 11°50′ E). This area mainly covers Sudan, South
Sudan, Eritrea, Djibouti, Somalia, Uganda, Kenya, and Tanzania. The
target region is characterized by widely diverse climates ranging
from tropical highlands to semi-arid and arid areas (Nicholson, 2001).
Although the focus of this study was on East Africa, Yemen was also
included as it is the origin of many lentil landraces.
Sowing date is known to affect plant development and growth
which, in turn, affects final yield (Bejiga, 1991). In northern India,
Singh and Saxena (1982) obtained the highest yield from lentil sown
in the first fortnight of November, while later sowing resulted in
lower yield. In Syria, seed yield was maximized by sowing in December, while delaying sowing to January and February reduced seed
yield by 25% (Saxena et al., 1983). In Ethiopia, Bejiga (1984, 1991)
found that early lentil sown between the last week of June and the
second week of July increased yield. Later sowing after mid-July
reduced both yield and time to maturity (Bejiga, 1991).
Matching the growth cycle of a crop to the rainfall environment is one of the most important factors affecting crop adaptation
to new areas. Differences in phenological development contribute
45–60% of the variation in seed yield in South Asia (Shrestha et al.,
2006; Siddique et al., 2003). Kusmenoglu and Muehlbauer (1998)
increased seed yield in lentil through the development of cultivars with shorter vegetative and flowering periods. Early flowering
is often thought an advantage in enabling long pod-filling period
before the onset of drought or high temperatures, during the later
stages of crop growth and development (Siddique et al., 2003; Silim
et al., 1993). For other conditions, where no serious water scarcity
is expected during the critical stages of crop growth and development, a different phenology might be required (Saxena, 2009; Solanki
et al., 2007).
Consequently, the first objective of this simulation study was to
determine the geographical areas in East Africa that offer potential for consistent lentil production. A mechanistic model was used
to calculate grain yields across 30 growing seasons of generated
weather data for each location on a 1 × 1 grid in regions that are
candidate areas for expansion of lentil production in East Africa.
The second objective of this study was to examine the change
in an agricultural management practice and a possible plant trait
to increase grain yield potential of lentil in East Africa. First, the effect
of a management practice was examined by exploring the effect of
different sowing dates on potential yield of lentil. Second, the effect
of phenology on yield potential was examined by comparing shortseason, early-flowering lentils to long-season, late-flowering ones.
The results of these simulations, therefore, offer information across
East Africa in guiding geographically based research in regard to
where lentil might be produced, and some research opportunities
to increase grain yield potential.
the SSM-Legumes model that demonstrated a robust predictive
capability in assessing variation in phenological development and
yield of lentil in a range of environments, with different rainfall patterns in the Middle East (Michel E. Ghanem, personal communication).
The SSM-Legumes model accounts for leaf area development as
a function of temperature that can be restricted by inadequate nitrogen and soil water. The leaf area index of the crop is used to
intercept solar radiation, which in turn is used to calculate crop
growth as a function of radiation use efficiency (Sinclair et al., 2014).
Radiation use efficiency, like leaf extension, is decreased proportional to soil water content (Soltani and Sinclair, 2012b). Thus, daily
growth was calculated based on incident solar radiation, leaf area
index, and soil water content. Finally, daily seed growth was calculated as a fraction of total dry matter, considering a linear increase
in harvest index.
One of the unique features of the SSM-model was that the daily
crop transpiration rate is intimately linked to daily crop growth. That
is, the transpiration rate is calculated as a function of crop growth
multiplied by the atmospheric vapor pressure deficit, divided by a
constant, mechanistically-based transpiration coefficient (Soltani and
Sinclair, 2012b). The parameters defining the transpiration coefficient have been shown to be stable across a wide range of conditions
within a crop species (Tanner and Sinclair, 1983). The transpiration rate was calculated each day so the soil water budget was
updated daily by water addition if there was rainfall and water
removal from the soil as a result of transpiration and direct evaporation from the soil surface.
Calculation of the soil water balance in the lentil model is similar
to the original SSM-Legumes of Soltani and Sinclair (2012b). That
is, the entire soil volume occupied by roots is considered as a single
compartment for calculating the reservoir for crop available soil
water. The initial depth of soil water extraction at plant emergence is set equal to 200 mm. Subsequently, the depth of extractable
soil water profile is increased steadily per biological day (defined
below). The final depth of soil water extraction is limited either by
phenological development, the maximum rooting depth capacity
of the crop, or by chemical and physical barriers in the soil (limited
soil depth). After beginning of seed growth (BSG), increases in rooting
depth were terminated.
Soil water status, calculated as the amount of transpirable soil
water (ATSW), was calculated daily using the water balance equation for an expanding volume of soil, corresponding to the addition
of the new soil layer explored by roots on that given day. Since lentil
is cultivated as a rainfed crop in East Africa and Yemen, no irrigation was scheduled in the simulations. Any new accessible water in
the soil due to root growth (NEWAT) is added to the total water volume
component together with daily rainfall (P, mm), and when relevant, irrigation input (I, mm). Water losses from run-off (R, mm), soil
evaporation (ES, mm), transpiration (TR, mm) and drainage (D, mm)
are subtracted from the soil water. Drainage water is calculated as
water that exceeds the total water holding capacity of the volume
of soil explored by roots. Run-off is calculated using a modified curve
technique as reported by Soltani and Sinclair (2012b).
2. Materials and methods
ATSWd = ATSWd−1 + Pd + NEWATd − Rd − ESd − TRd − Dd
2.1. Model description
In addition, a new algorithm that accounts for crop survival at
very low soil water content (Michel E. Ghanem, personal communication) was added to the lentil model based on concepts developed
by Sinclair (2000). In low-yielding environments, like East Africa,
plant survival under severe drought is crucial for crop performance. The survival phase is characterized by stomatal closure so
that continued water loss is considered to be through the entire epidermis based on leaf epidermal conductance (EPCOND). As water
is lost from the leaves, leaf relative water content decreases until
it reaches a lethal leaf relative water content (LTLRWC). Leaf area
Soltani and Sinclair (2012b) recently presented a non-calibrated,
mechanistic, comparatively simple model called Simple Simulation Model (SSM) for simulating growth and yield of crops. The
robustness of an SSM-Legumes version of the model, generic to grain
legume species, has been demonstrated in several studies over a
wide range of environments for various legume species including
soybean (Sinclair et al., 2010), chickpea (Vadez et al., 2013), and bean
(Marrou et al., 2014). Recently, we parameterized a lentil version of
(1)
26
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
was decreased to account for the water loss in reaching LTLRWC for
successive leaves of the crop canopy. For lentil, the values of these
two physiological input parameters of EPCOND and LTLRWC were
set at 0.1 mm s−1 and 0.355 g g−1, respectively (Leport et al., 1998;
Turner et al., 2001).
The duration of each development stage of the plants was defined
by the biological days required for completion of the phenological
stage (Soltani and Sinclair, 2012b). Biological days represent the total
number of days required to complete a development stage when
plants are grown under optimum photoperiod and temperature conditions. The actual number of days to complete a development stage
is greater than the biological days depending on whether actual daily
temperature and photoperiod are limiting or not. Therefore, average
daily temperature and daily photoperiod were crucial in the model
to account for their influence on crop development. In the model,
biological days are defined for the following stages: sowing to emergence, emergence to flowering (stage R1 in lentil), R1 to R3
(beginning of pod formation), R3 to R5 (beginning of seed fill), R5
to R7 (end of seed fill) and R7 to R8 (maturity).
2.2. Model entries: weather generation
The model requires daily input of minimum and maximum temperature, solar radiation, and rainfall. These data were obtained from
the weather generator described below. The weighted daily vapor
pressure deficit, which is required in the calculation of transpiration rate, was calculated from minimum and maximum temperature
based on the approach suggested by Tanner and Sinclair (1983).
To understand the variability in the range of crop response to
weather, it is necessary to simulate crop yield for at least 20 growing
seasons at each location (Sinclair et al., 2014). Given that reliable
and complete long-term weather datasets are scarce in Africa, it was
necessary to generate weather data. The National Centers for Environmental Prediction (NCEP) and National Center for Atmospheric
Research (NCAR) re-analysis baseline data were used to produce a
retroactive record of daily data (Dee et al., 2011; Uppala et al., 2005).
The surface grids were analyzed in the Weather Research and Forecasting (WRF) tool to extract daily values for the last 30 years at
the centroid of a 1° × 1° point grids in each of the regional blocks.
Given the importance of these datasets in the crop simulations, it was necessary to verify whether they do, in fact, represent
the patterns of rainfall and temperature distribution. The amount
and distribution of precipitation and temperature are among the
most important environmental variables that influence the soil water
balance, crop growth and other processes. Therefore, an independent comparison was performed for a few locations of similar
latitudes, for which historical data were available. The records were
checked not only for distribution of precipitation but also for
maximum and minimum temperature in order to detect obvious
errors. Simulated values were shown to follow the same distribution as the observed ones although the absolute values were not
necessarily exactly the same (Ghanem et al., unpublished results).
2.3. Model parameters: soil and crop parameters
Simulation were run using fixed sowing date options with no
irrigation. Since no extensive soil data were available, in all locations the volumetric transpirable soil water was set equal to 0.13 m3
m−3 and the maximum rooting depth was set equal to 1500 mm.
Of course, these two parameters can be varied if such information
is available for a specific location. The main crop parameters were
extracted from previous published studies or were experimentally measured, and are reported in Table 1.
The same temperature function was used in all simulations with
the base temperature equal to 2 °C, an optimum temperature of 20 °C
and a ceiling temperature equal to 30 °C. The photoperiod response
Table 1
Crop parameters for lentil, as needed by the SSM-legumes model. Parameter values
have been calculated based on experimental results or estimated based on similarities with other grain legumes. Parameters were previously reported in Michel
E. Ghanem, (personal communication).
Parameter
Value
Reference
Phyllochron
50.0 °C/leaf
A coefficient (exponent) in power
relationship between plant leaf
area and mainstem node
number
Specific leaf area
Potential radiation use efficiency
Harvest index increase (PDHI)
2.1
Michel E. Ghanem,
personal communication
Michel E. Ghanem,
personal communication
0.0069 m2 g−1
2.05 g MJ−1
0.018 d−1
Transpiration coefficient
0.5 Pa
FTSW threshold when dry matter
production starts to decline
(WSG)
FTSW threshold when leaf area
development starts to decline
(WSSL)
FTSW threshold when N2 fixation
starts to decline (WSSN)
N content in green leaves
N content in senescent leaves
N content in green stems
N content in senescent stems
N content in grain
Maximum N uptake through
fixation
0.47
McKenzie and Hill, 1989
McKenzie, 1987
Michel E. Ghanem,
personal communication
and references therein
cited
Michel E. Ghanem,
personal communication
and references therein
cited
Michel E. Ghanem,
personal communication
0.55
Michel E. Ghanem,
personal communication
0.30
Michel E. Ghanem,
personal communication
McKenzie and Hill, 1989
Whitehead et al., 2000
Ayaz et al., 2004
Michel E. Ghanem,
personal communication
2.6 g m−2
1.0 g m−2
0.019 g·g−1
0.03 g·g−1
0.043 g.g−1
0.50 g.m−2.d−1
was defined by two variables; the critical photoperiod, which is 11 h
for lentil and the photoperiod sensitivity ppsen, which is 0.28 for
lentil (Michel E. Ghanem, personal communication).
2.4. Crop yield simulations
2.4.1. Sowing date and plant density
In these simulations, determination of sowing dates were not
based on a threshold of accumulated soil water. Instead, simulations were done over a range of defined sowing dates from day 152
(1st of June) till day 264 (21st of September) at 1-wk intervals.
Sowing density was fixed at 60 plants m−2, which corresponds to
the sowing density used by farmers in Ethiopia (Bejiga, 1991).
2.4.2. Phenological parameters
In lentil, genotypes are classified by maturity groups reflecting
their adaptability to various latitudinal zones as a result of sensitivity in development to temperature. The expectation was that earlyflowering lentil genotypes are often thought to have an advantage
in comparison with late-flowering genotypes. While an optimization could be attempted for each phenological cycle, for these initial
simulations a single set of development parameters for each of the
long-season (late emergence-late flowering) and short-season lentil
(early emergence-early flowering) genotypes was explored. The differences in development of the early- and long-season genotypes
were defined by the biological days of each development stage
(Table 2). Initial simulations were done for short-season, earlyflowering, lentil genotypes similar to the ones cultivated in the
Middle East and North Africa (Siddique et al., 1998; Thomson et al.,
1997). Subsequently, simulations were done for long-season, lateflowering genotypes that are grown in Southern latitudes like
Ethiopia and South Asia (Bejiga et al., 1996; Mondal et al., 2013).
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
Table 2
Developmental parameters for simulation of short-cycle and long-cycle lentil in East
Africa including the biological days required for each of the simulated phenological stages.
Parameter
Long-cycle cultivar
(Siddique et al., 1998;
Thomson et al., 1997)
Short-cycle cultivar
(Bejiga et al., 1996;
Mondal et al., 2013)
Critical photoperiod (h)
Photoperiod sensitivity
11
0.28
Biological days
11.0
43.0
5.7
13.0
16.0
14.0
11
0.28
Sowing to emergence
Emergence – R1
R1 – R3
R3 – R5
R5 – R7
R7 – R8
7.8
30.0
5.7
13.0
16.0
14.0
The two plant types differed mainly in their time from sowing to
emergence and emergence to flowering (stage R1 in lentil) (Table 2).
3. Results
3.1. Rainfall spatial distribution
Eastern Africa’s diverse topography and large water bodies such
as the Indian Ocean and Lake Victoria contribute to a high variance in spatial distribution of rainfall. Fig. 1 shows variation in yearly
precipitation over the region. Ethiopian highlands, Uganda, as well
as southern parts of South Sudan, and eastern parts of Tanzania show
the highest annual rainfall: between 1500 and 2000 mm/year (Fig. 1).
The Greater Horn of Africa, Somalia, Kenya, as well as eastern parts
of Ethiopia and western parts of Tanzania show lower cumulative
rain. Most parts of Sudan and northern parts of South Sudan are a
very arid zone with an annual precipitation less than 200 mm per
year (Fig. 1).
East Africa can be divided into four more or less homogenous
rainfall regions (Fig. 1): Region A (Greater Horn of Africa); Region
B (Ethiopian Highlands); Region C (Central East Africa); and Region
D (Southern East Africa). The annual cycle of rainfall in East Africa
is generally characterized by two rainy seasons: the “long rains” in
March-May (MAM) and the “short rains” which typically start midOctober and continue until mid- December (OND) (Black et al., 2003;
Clark et al., 2003). In general, East African climate is normally dry
between the “long and short rains” seasons. The exception to this
is in the high plateau regions in the northern Great Rift Valley (central
and northern Ethiopia) (Fig. 1, Region B), which has a third and stronger rainy season (Kiremt) during June to September (Segele and
Lamb, 2005).
3.2. Baseline simulated yield and production risk
The initial objective of this simulation study was to assess geographical viability of lentil production in East Africa. The model,
which included the new algorithm accounting for crop survival at
very low soil water content, was run for all locations and years using
the baseline parameters, phenology, and sowing date most commonly found in the literature for lentil in Ethiopia, i.e. a shortseason phenology (parameters in Table 2) sown in June on day of
the year 152 (Fig. 2). As shown in Fig. 2B, the Ethiopian highlands
were simulated to have the highest average grain yields ranging
between 1.2 to 2.0 t ha−1 (120 and 200 g m−2) dry weight. Portions
of south Uganda, north eastern Tanzania and a pocket in south Kenya
and Tanzania were simulated to have average grain yields ranging
between 1.2 to 1.6 t ha−1 dry weight. South Sudan and most of Tanzania had a simulated average yield ranging between 0.8 to 1.2 t ha−1.
Many of the locations in the Great Horn of Africa (Somalia and
27
eastern Ethiopia) had a low simulated average yield ranging between
0.4 to 0.8 t ha−1. Most of the locations in Sudan and Yemen had a
very low simulated average yield (less than 0.4 t ha−1). There were
some locations in the northern parts of Sudan where the lack of rainfall resulted in no simulated crop growth (i.e., no germination or
early crop termination every year) (Fig. 2B).
The risk of being unable to obtain yield (expressed as number
of years where yield could not be obtained) can be a stronger criteria for acceptability for industrial and commercial development
of a crop in a region than average yield alone. Setting this threshold at two or more years of failure out of 30, large areas of East Africa
appear not to be suitable to lentil growth under the prevailing current
cropping practices (i.e., a short-season phenology crop sown in June).
As shown in Fig. 2A, lentil production is not appropriate in a large
band in the northern tier of East Africa (Sudan, Eritrea, Yemen, and
northern Ethiopia). In addition, much of the Great Horn of Africa
region (Somalia and eastern Ethiopia) and large parts of western
Tanzania are not appropriate for lentil production. Even with a more
relaxed threshold criterion at three or more years of failure out of
the 30 growing seasons, there are still large areas in Tanzania,
Somalia, Uganda, and Kenya that seem unlikely to be suited for consistent production of lentil (Fig. 2) under the simulated conditions.
On the contrary, Ethiopian highlands, South Sudan, eastern Kenya
and Tanzania appear suitable for lentil production (Fig. 2A).
3.3. Change in sowing date
One management practice for potential increase in lentil yield
is to change sowing date. Farmers in Amhara and Oromia regions
in Ethiopia usually sow lentil in June to July on well-drained soil,
while farmers of waterlogged black soil can sow as late as the end
of September (Bejiga, 1991). Simulations were done over a range
of sowing windows from day of year (doy) 152 (1 June) until 264
doy (21 September) at 1-week intervals. The sowing date of early
June (152 doy) was taken as the basis for assessing the impact of
the change in sowing date on potential yield.
Neither the absolute values nor the distribution pattern of simulated average grain yield changed when sowing date was shifted to
early July (182 doy). Of the remaining dates, the results for 201, 229,
243 and 264 doy are presented to illustrate the full range of results.
The probabilities of yield increase as a result of the delaying sowing
date are shown in Fig. 3. Except for a very few locations in northern
Sudan, delaying sowing until late July (201 doy) resulted in a higher
probability of average yield increase for many simulated locations
in East Africa (Fig. 3A). The most consistently large yield increases
with delayed sowing were in the central portion of East Africa, especially in Ethiopia, South Sudan, Uganda, and some of the areas of
in central and southern Somalia, where most of the locations had
more than 85% probability of yield gain. On the other hand, there
were locations between these two areas and around them (like in
Tanzania and Sudan), where the probability of yield increase was
less than 55%, i.e. there was a likelihood of a yield decrease (Fig. 3A).
A similar pattern was observed when sowing was delayed until midAugust (229 doy) (Fig. 3B). For this date, the highest probability of
yield gain (more than 85%) was extended eastwards in Ethiopia and
Somalia. Delaying sowing to September (both early (243 doy) and
late (264 doy) ) (Fig. 3C and D) showed the highest probability of
yield gain (85% increase) in Kenya, Uganda, northern parts of Tanzania and central parts of Somalia (Fig. 3C and D).
3.4. Change in phenology
The standard lentil model simulated yield based on the definition of the development stages of a short-season lentil genotype
(early emergence-early flowering). Simulations were done, for each
sowing date, for an altered late-maturity long-season genotype (late
28
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
Fig. 1. Annual rainfall distribution at simulated locations in East Africa.
emergence-late flowering) in which the number of days to emergence and the number of days from emergence to flowering were
greater (Table 2). Fig. 4 shows the probabilities of yield increase as
a result of a long-season genotype (201, 229, 243, and 264 doy).
A late-emerging, late-flowering crop sown early June (152 doy)
had a positive impact on yield in nearly all locations in South Sudan
and Ethiopia, Uganda as well as southern parts of Somalia and coastal
areas of Kenya (Fig. 4A). On the contrary, in most parts of Sudan,
Yemen and Tanzania, the probability of yield increase by prolonging
the crop cycle was less than 55%. Further delaying the sowing
until July (201 doy) or mid-August (229 doy) for a long-season genotype of lentil restricted the extent of the area where the highest
probability of yield increase (85%) was expected to the Ethiopian highlands, South Sudan and Uganda (Fig. 4B and C) compared to the
baseline scenario of sowing at day 152 with the same phenology.
When sowing was performed later in September (243 or 264 doy),
the area where the highest probability of yield increase moved further
to the south of East Africa covering only a few locations of southern
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
29
Fig. 2. Average grain yield (WGRN, g dry weight m-2) (A) and number of years without yield (B) for standard short cycle lentil simulated at each location for a fixed sowing
date at 152 day-of-year (doy).
30
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
Fig. 3. Probability of grain yield increase of simulating different sowing day-of-year (SD) (A) 201, (B) 229, (C) 243, and (D) 264 as compared to a sowing date of 152 doy.
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
31
Fig. 4. Probability of grain yield increase of simulating long-cycle lentil as compared to a short-cycle lentil for different sowing day-of-year (SD) (A) 152, (B) 201, (C) 229,
(D) 243, and (E) 264.
32
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
Fig. 5. Simulation of actual transpirable soil water (ATSW) in mm of water left in the soil after harvest for short-cycle and long-cyclel lentil sown on different sowing dayof-year (SD) (A and B) 152, (C and D) 201, (E and F) 229, (G and H) 243, (I and J) 264.
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
Fig. 5. (continued)
33
34
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
parts of Ethiopia and South Sudan but almost entirely Uganda and
Kenya as well as the northern locations in Tanzania (Fig. 4D and E).
3.5. Soil water content at the end of the crop cycle
The amount of actual transpirable soil water (ATSW, mm) that
was in the root zone at the end of the crop cycle was examined as
an indicator of the extent of water use by the crop. A low value of
ATSW indicates that the crop cycle was better synchronized with
rainfall which allowed more effective use of the available water. Fig. 5
represents the ATSW for both the long-season (Fig. 5A, C, E, G, I)
and short-season (Fig. 5B, D, F, H, J) lentil genotypes for each of the
sowing dates. For all sowing dates, northern locations of Sudan
showed the lowest ATSW values (zero) especially on the earliest
sowing date (152 doy) for both phenologies (Fig. 5A, B). For a crop
sown in June (152 doy), the highest simulated ATSW values were
found in Ethiopia, South Sudan and Uganda (more than 120 mm of
water) for both phenologies. For the same sowing date, low simulated ATSW values (less than 50 mm) were shown for most parts
of Sudan, the Great Horn of Africa, Kenya and Tanzania for shortseason phenology.
Later sowing dates moved the area where simulated ATSW was
the highest southwards toward Kenya, Uganda and Tanzania (Fig. 5).
In general, the same spatial pattern of ATSW is observed for the same
sowing date, on long- and short-season genotypes. However, for the
highest simulated ATSW values, long-season lentil genotypes left
less water in the soil (lower ATSW) than the short-season genotypes, especially for sowing dates following 201 doy (Fig. 5C–J).
4. Discussion
4.1. Areas for lentil production
The SSM-legume model used in this assessment was based on
the parameters presented for lentil by (Michel E. Ghanem, personal communication). The simulation results presented here using the
baseline parameters, phenology and sowing date most commonly
found in the literature for lentil in southern latitudes in East Africa
showed that the model was able to predict regions of higher yield
potential. The simulations predicted the highest yield potential for
lentil in the Ethiopian highlands (Fig. 1). This area of higher yield
coincides with Amhara, Oromia, Tigray and the “Southern Nations
Nationalities and Peoples” (SNNP) regions, known to be the main
lentil producing regions in Ethiopia (Regassa et al., 2006).
The average observed seed yield of lentil in farmers’ field in Ethiopia is generally about 0.6–0.8 t ha−1 (Regassa et al., 2006). Not
surprisingly, the simulation results in this study predicted a higher
maximum average grain yield for the same areas with yield ranging
between 1.6 and 2.0 t·ha−1 (Fig. 1). Yields of about 2.0 t ha−1 have
been reported from experiments performed under controlled field
experimental conditions in Ethiopia (Erkossa et al., 2006).
Lentil is dominantly produced by smallholder farmers and most
of them follow a traditional farming system based on indigenous
knowledge. Risk decisions by farmers in regard to the introduction of a change in management practices or a new plant trait may
cause the probability of yield increase to be at least as important
as the long-term average yield increase. Very likely, risk-averse
farmers may want to avoid the changes that result in a substantial
fraction of the growing season with yield decreases even if the longterm average yield is higher. Since in one location nearly all
management practices or traits can result in a given season, either
an increase or a decrease in yield depending on the weather of that
individual growing season, it is important for crop breeders and
farmers to understand the probability of yield increase (Sinclair et al.,
2014).
The inability to obtain yield in two or more growing seasons out
of 30 was a criterion used to identify the geographical areas in East
Africa for consistent low-risk, lentil production. Simulation results
showed under the current conditions of short-season genotypes
sown in June that large areas of East Africa appear not to be suitable for lentil production. Fig. 2A, shows that large parts of the
northern tier of East Africa and much of the Great Horn of Africa,
as well as western Tanzania are high risk areas for lentil production. On the contrary, the Ethiopian highlands, South Sudan, Uganda,
eastern Kenya and Tanzania appear suitable for lentil production
(Fig. 2A). Lentil is not currently a crop in Kenya, Uganda and Tanzania (Bejiga and Degago, 2000). However, the simulation results
showed that even under the actual parameters, phenology, and
sowing date most commonly used in southern latitudes, lentil can
potentially become a crop in large areas in these countries.
4.2. Sensitivity to sowing date
Changing sowing dates could be an option for increasing yield
in future climates (Ludwig and Asseng, 2010). In general, delaying
sowing until mid-August (229 doy) resulted in a higher probability of average yield increase across many simulated locations in
central East Africa covering the countries of Ethiopia, South Sudan
and Uganda (Fig. 3). The probability of yield increase as a result of
delayed sowing was simulated to always be higher than 85% (for
all dates) for Uganda. In the northern tier of East Africa, except a
small pocket in Sudan, delaying sowing beyond mid-August does
not seem useful as it did not improve the probability of yield increase.
Delaying sowing to September (243 or 264 doy) shifted the
highest simulated yield gain (85% increase) to the south toward
Kenya, Uganda, northern parts of Tanzania and central parts of
Somalia (Fig. 3C and D). This coincides with the end of the rainy
season in this area and would therefore offer a potential of growing
lentil in this area by replacing the fallow. This would not compete
with other crops like maize, which are grown during the rainy season
in this area of East Africa. Of course, this hypothesis must be explored experimentally. Shifting sowing dates until the end of
September (264 doy) is likely not to be useful in Ethiopia (Fig. 3D).
These results contrast with common advocacy of early sowing
as a means to avoid rising temperatures and drought during the reproductive phase and maximize lentil yield. With a defined growing
season, the early sowing of lentil is reported to produce the highest
seed yield in Mediterranean type of environments such as Italy, Syria,
Western Australia (Materne et al., 2007). Under these conditions,
an early sowing lentil crop makes better use of seasonal precipitation than the crop from a later date of sowing (Saxena et al., 1983).
However, when lentil is sown early, depending on the performance of varieties and soil type, increased weed competition, biotic
stresses, and waterlogging incidences may occur and affect its productivity (Materne et al., 2007). Late sowing in the Middle-East is
reported to alter the vegetative and phenological development of
the crop through changes in temperature and photoperiod (Erskine
and Saxena, 1993; Erskine et al., 1994).
In Ethiopia, lentil is predominantly grown during the main rainy
season (Kiremt) (June to September). The simulation results for Ethiopia contrast with some experimental conclusions from that of late
June to mid-July sowing in both mid- to high-altitude areas (Bejiga,
1991; Regassa et al., 2006). Bejiga et al. (1996) reported that later
sowing after mid-July reduced yield in Ethiopia. However, there was
a large yield variation in yield in the data presented by Bejiga et al.
(1996), highlighting the challenge in using conventional experimentation conducted over limited number of seasons (five in the
study of Bejiga et al., 1996) to screen for an appropriate sowing date
in a location (Sinclair and Muchow, 2001). In chickpea, Soltani and
Sinclair (2012a) reported also that calculations of mean yield based
on 10-year periods did not match the overall simulated long-term
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
35
mean in many cases, indicating that even ten years of experimentation may be insufficient to reflect the overall crop behavior in a
location.
4.3. Sensitivity to phenology parameters
Matching the phenology of a crop to an environment is key to
improving adaptation and increasing crop yield. The simulations
results showed that lengthening the time to flowering and crop
growing cycle was very similar to the distribution resulting from
the change in sowing date. That is, many locations had probabilities of yield increase greater than 70% in nearly all locations in South
Sudan and Ethiopia, and Uganda for all sowing dates (Fig. 4).
However, many locations in Sudan, Yemen and Tanzania showed
only a small benefit from increasing the duration of the crop season
(less than 55%) for almost all sowing dates. Like the distribution of
probability of yield gain resulting from a late sowing alone (Fig. 3D),
a combination of late sowing (September 243 or 264 doy) and
long-season genotypes shifted the area of highest probability of yield
increase to the south of East Africa toward Uganda and Kenya as
well as northern locations of Tanzania (Fig. 4D and E).
These simulation results generally indicated higher yield with
a long-season crop. These results are consistent with some experimental results in Ethiopia, reporting that late-maturing genotypes
were found to be favored when rainfall occurred over a longer period
(Bejiga et al., 1995). In Australia, late sowing improved the relative performance of the late-flowering genotypes that perform poorly
in drought years (Materne, 2003).
Conversely, these results are in contradiction with experimental findings on lentil grown in the dry areas of Syria where the
highest yielding lentil genotypes produced a large amount of mass,
flowered early, and had a brief, rapid seed-filling phase (Silim et al.,
1993). Time to flowering accounted for 49% of the variation in the
seed yield, indicating that drought avoidance through early flowering was key to minimize the effects of drought stress (Silim et al.,
1993). The opposite was observed for the same genotypes in wetter
years. Similarly, early flowering and maturing genotypes were high
yielding at low-rainfall, low-yielding environments in Australia but
the yields of these genotypes were low compared to the best
medium-rainfall, mid-flowering genotypes over eight locations and
five years (Materne, 2003). A possible explanation for these discrepancies between sites would be that, although early flowering
is important for drought escape, it is not effective if genotypes have
a relatively low mean yield over many seasons and sites, due to an
inability to respond to increasing available soil moisture (Materne,
2003; Materne and Siddique, 2009; Turner et al., 2001). Additionally, avoidance may not be the only mechanism for drought tolerance
in lentil (Leport et al., 1998; Shrestha et al., 2006).
Comparing the water available in the soil at crop maturity by
examining the ATSW (Fig. 5) clearly shows the deficiency of the
short-season phenology, compared to the long-season one. Fig. 5
shows that for most sowing dates (except 152 doy) short-season
lentil genotypes left more water in the soil (higher ATSW values)
than the long-season genotypes. This indicates that for areas of higher
probability of yield increase (located in the north and center of East
Africa for early sowing and in the southern locations of East Africa
for later sowing), there is still water in the soil that can be used as
a resource to maximize mass accumulation and yield formation.
Short-season phenology shows a disadvantage compared to longseason phenology in the potential use of the water resource.
4.4. Combining changes in sowing date and phenology
Generally, delaying sowing and a longer phenology tended to have
similar distribution patterns of probability of yield increase in East
Africa. However, the relative value of the change in management
Fig. 6. Simulation of average yield gain (WGRN, g dry weight m-2) at each grid location by changing (A) sowing date, (B) phenology, and (C) changing both sowing
date and phenology.
36
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
Fig. 7. Simulation of optimum sowing window for long-cycle lentil genotypes.
practice and/or trait differed by locations. As shown in Fig. 6, the
simulated average yield gain that can be reached by changing sowing
date alone (Fig. 6A) is less than the yield gain through lengthening lentil phenology (Fig. 6B). This can been explained by the fact
that, in very dry areas, shifting the cycle position, without substantially changing cycle duration, won’t allow escape of extreme drought
and subsequent significant yield reduction, or early crop termination. In the case of changing sowing date, major gains (above
50 g m−2) were mainly limited to Ethiopia and eastern Tanzania, while
changing sowing date was unlikely to be beneficial in southern
Tanzania (Fig. 6A). Under the effect of a longer-season genotype, yield
gain was further extended (Fig. 6B) to many locations of Ethiopia
and Uganda and some locations in South Sudan. However, the
maximum extent of average yield gain was reached by combining
changes in both sowing date and a longer phenology to cover Ethiopia, South Sudan, Uganda and Kenya, as well as some parts of
Tanzania and Somalia (Fig. 6C).
The simulation results allowed a determination of the optimum
sowing windows for each location for long-season genotypes of lentil.
Fig. 7 represents the optimum sowing windows when average grain
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
yield was simulated highest across the 30 growing seasons. For longseason genotypes of lentil, an optimum sowing window was found
between June and July (152–229 doy) for areas to the north of the
Rift Valley (Fig. 7). Indeed, in the Northern part of Eastern Africa
there is limited opportunity for the crop to benefit from rainfall in
the second rainy season since the dry season is very long. Except
for some locations in Tanzania, later sowing dates (229–243 doy)
were found to be optimal in southern areas of East Africa. This is
consistent with the high possibility for the crop to capture water
from the second rainy season if they are not sown too early. In Tanzania, substantial rainfall is commonly experienced from the end
of September.
5. Conclusions
The results of these simulations offer information across East
Africa in guiding geographically based research in regard to where
lentil might be produced, and some research opportunities to increase grain yield potential that can be explored. The results of these
simulations offered several crucial insights into developing lentil production in East Africa:
1 One of the key findings of this study is that there is a potential
to further expand the geographical area in which lentil is currently grown in East Africa into Uganda, Kenya, Tanzania and even
Somalia. Obviously, local experiments are needed to evaluate the
possibility of growing lentils and evaluating the genetic material available.
2 Delaying sowing alone or in combination with long-phenology
genotypes can result in a higher probability of crop yield increase in lentil producing areas of East Africa. However, the
benefit of the change in this management practice or trait needs
to be evaluated both by the increase in average yield and by probability of yield increase.
3 These simulations show that response to management practice or trait modification in a breeding effort strongly depends
on the location of deployment. Certainly, these results do not
support the concept of cultivar development for “wide
adaptation”.
Acknowledgment
This work was supported by the CGIAR Research Program on
Grain Legumes and the USAID/CGIAR-US Universities Linkage
Program.
References
Ayaz, S., McKenzie, B.A., Hill, G.D., McNeil, D.L., 2004. Variability in yield of four grain
legume species in a subhumid temperate environment. II. Yield components.
J. Agric. Sci. 142, 21–28.
Bejiga, G., 1984. Lentils in Ethiopia. Lens 11, 1–6.
Bejiga, G., 1991. Effect of sowing date on the yield of lentil (Lens culinaris Medik.).
J. Agron. Crop Sci. 167, 135–140.
Bejiga, G., Degago, Y., 2000. Region 4: Sub-Sahara Africa. In: Knight, R. (Ed.), Linking
Research and Marketing Opportunities for Pulses in the 21st Century.
Bejiga, G., Tsegaye, S., Tullu, A., 1995. Stability of seed yield for some varieties of lentil
grown in the Ethiopian highlands. Crop Res. 9, 337–343.
Bejiga, G., Tsegaye, S., Tullu, A., Erskine, W., 1996. Quantitative evaluation of Ethiopian
landraces of lentil (Lens culinaris). Genet. Res. Crop Evol. 43, 293–301.
Black, E., Slingo, J., Sperber, K.R., 2003. An observational study of the relationship
between excessively strong short rains in coastal East Africa and Indian Ocean
SST. Mon. Weather Rev. 131, 74–94.
Chen, W., Sharma, H.C., Muehlbauer, F.J., 2011. In Crop Protection Compendium for
Chickpea and Lentil Diseases and Pests. The American Phytopathological Society,
St Paul, MN, p. 166.
Clark, C.O., Webster, P.J., Cole, J.E., 2003. Interdecadal variability of the relationship
between the Indian Ocean zonal mode and East African coastal rainfall anomalies.
J. Climate. 16, 548–554.
37
Dee, D.P., Uppala, S.M., Simmons, J., Berrisford, P., Poli, P., Kobayashi, S., et al., 2011.
The ERA-Interim reanalysis: configuration and performance of the data
assimilation system. Q. J. R. Meteorol. Soc. 137, 553–597.
Erkossa, T., Stahr, K., Gaiser, T., 2006. Soil tillage and crop productivity on a Vertisol
in Ethiopian highlands. Soil Tillage Res. 85, 200–211.
Erskine, W., Saxena, M.C. (Eds.), 1993. Lentil in South Asia. In: Proceedings of the
Seminar on Lentils in South Asia. ICARDA, Aleppo, Syria.
Erskine, W., Tufail, M., Russell, A., Tyagi, M.C., Rahman, M.M., Saxena, M.C., 1994.
Current and future strategies in breeding lentil for resistance to biotic and abiotic
stresses. Euphytica 73, 127–135.
FAOSTAT, 2011. Food and Agriculture Organization of the United Nations. FAO Stat
statistical database. <http://faostat.fao.org>.
Kelly, J.D. 2004. Advances in common bean improvement: some case histories with
broader applications. In: McCreight, J.D., Ryder, E.J. (Eds.), Advances in Vegetable
Breeding, Proceedings of the 26th International Horticultural Congress, Toronto,
Canada, pp. 99–122.
Kusmenoglu, I., Muehlbauer, F.J., 1998. Genetic variation for biomass and residue
production in Lentil: I. Relation to agronomic traits. Crop Sci. 38, 907–910.
Ladzinsky, G., Smartt, J., 2000. Opportunities for improved adaptation via further
domestication. In: Knight, R. (Ed.), Linking Research and Marketing Opportunities
for Pulses in the 21st Century. pp. 257–263.
Leport, L., Turner, N.C., French, R.J., Tennant, D., Thomson, B.D., Siddique, K.H.M., 1998.
Water relations, gas partitioning and redistribution of dry matter from exchange,
and growth of cool-season grain legumes in a Mediterranean-type environment.
Eur. J. Agron. 9, 295–303.
Ludwig, F., Asseng, S., 2010. Potential benefits of early vigor and changes in phenology
in wheat to adapt to warmer and drier climates. Agric. Syst. 103, 127–136.
Marrou, H., Sinclair, T.R., Metral, R., 2014. Assessment of irrigation scenarios to
improve performances of lingot bean (Phaseolus vulgaris) in the South West of
France. Eur. J. Agron. 59, 22–28.
Materne, M., 2003. Importance of phenology and other key factors in improving the
adaptation of lentil (Lens culinaris Medikus) in Australia. Ph.D. thesis. The
University of Western Australia, Perth, Western Australia.
Materne, M., Siddique, K.H.M., 2009. Agro-ecology and crop adaptation. In: The Lentil
Botany, Production and Uses. CABI International, UK.
Materne, M., McNeil, D., Hobson, K., Ford, R., 2007. Abiotic stresses. In: Yadav, S.S.,
McNeil, D.L., Stevenson, P.C. (Eds.), Lentil: An Ancient Crop for Modern Times.
Springer, Dordrecht, The Netherlands, pp. 315–329.
McKenzie, B.A., 1987. The Growth, Development and Water Use of Lentils (Lens
culinaris Medik.). Ph.D. thesis. Lincoln University, New Zealand.
McKenzie, B.A., Hill, G.D., 1989. Environmental control of lentil (Lens culinaris) crop
development. J. Agric. Sci. 113, 62–72.
Mondal, M.M.A., Puteh, A.B., Abdul Malek, M., Roy, S., Rafii Yusop, M., 2013.
Contribution of morpho-physiological traits on yield of lentil (Lens culinaris
Medik). Aust. J. Crop Sci. 7, 1167–1172.
Nicholson, S.E., 2001. A review of climate dynamics and climate variability in Eastern
Africa. In: The Limnology, Climatology and Paleoclimatology of the East African
Lakes, pp. 25–56.
Pundir, R.P.S., Nyange, L.R., Sambai, L.M., 1996. Bright prospects for chickpea
cultivation in Tanzania. Int. Chickpea Newsl. 3, 11–12.
Regassa, S., Dadi, L., Mitiku, D., Fikre, A., Aw-Hassan, A., 2006. Impact of technologies
in selected lentil growing areas of Ethiopia: EIAR Research report number 67.
Saxena, M.C., 2009. Plant morphology, anatomy and growth habit. In: Erskine, W.,
Maeuhlbauer, F., Sarker, A., Sharma, B. (Eds.), The Lentil: Botany, Production and
Uses, 1st ed. CABI Publishing, London, UK, pp. 34–46.
Saxena, M.C., Murinda, M.V., Turk, M., Trabulsi, N., 1983. Productivity and water-use
of lentil as affected by date of sowing. Lens 10, 28–29.
Segele, Z.T., Lamb, P.J., 2005. Characterization and variability of Kiremt rainy season
over Ethiopia. Meteorol. Atmos. Phys. 89, 153–180.
Shrestha, R., Turner, N.C., Siddique, K.H.M., Turner, D.W., 2006. Physiological and yield
responses to water deficits among lentil genotypes from diverse origins. Aust.
J. Agric. Res. 57, 903–915.
Siddiq, M., Uebersax, M.A., 2012. Dry Beans and Pulses: Production, Processing and
Nutrition. John Wiley & Sons, p. 408.
Siddique, K.H.M., Loss, S.P., Regan, K.L., Pritchard, D.L., 1998. Adaptation of lentil (Lens
culinaris Medik) to short season Mediterranean-type environments: response
to sowing rates. Aust. J. Agric. Res. 49, 1057–1066.
Siddique, K.H.M., Loss, S.P., Thomson, B.D., 2003. Cool Season Grain Legumes in
Dryland Mediterranean Environments of Western Australia: Significance of Early
Flowering in Management of Agricultural Drought – Agronomic and Genetic
Options. Oxford University Press, New Delhi, India.
Silim, S.N., Saxena, M.C., Erskine, W., 1993. Adaptation of lentil to the Mediterranean
environment. 2. Response to moisture supply. Exp. Agric. 29, 21–28.
Sinclair, T.R., 2000. Model analysis of plant traits leading to prolonged crop survival
during severe drought. Field Crops Res. 68, 211–217.
Sinclair, T.R., Muchow, R.C., 2001. System analysis of plant traits to increase grain
yield on limited water supplies. Agron. J. 93, 263–270.
Sinclair, T.R., Messina, C.D., Beatty, A., Samples, M., 2010. Assessment across the
United States of the benefits of altered soybean drought traits. Agron. J. 102,
475–482.
Sinclair, T.R., Marrou, H., Soltani, A., Vadez, V., Chandolu, K.C., 2014. Soybean
production potential in Africa. Global Food Secur. 3, 31–40.
Singh, H.P., Saxena, M.C., 1982. Response of lentil genotypes to date of planting. Lens
9, 30–31.
Solanki, I.S., Yadav, S.S., Bahl, P.S., 2007. Varietal adaptation, participatory breeding
and plant type. In: Yadav, S.S., McNeil, D.L., Stevenson, P.C. (Eds.), Lentil: An
38
M.E. Ghanem et al./Agricultural Systems 137 (2015) 24–38
Ancient Crop for Modern Times. Springer, Dordrecht, The Netherlands, pp.
255–274.
Solh, M.B., 1996. A model for technology transfer, NVRP/ICARDA.
Soltani, A., Sinclair, T.R., 2012a. Optimizing chickpea phenology to available water
under current and future climates. Eur. J. Agron. 38, 22–31.
Soltani, A., Sinclair, T.R., 2012b. CABI (Ed.), Modeling Physiology of Crop Development,
Growth and Yield. Wallingford, Oxfordshire, UK.
Tanner, C.B., Sinclair, T.R., 1983. Efficient water use in crop production: research or
re-search? In: Taylor, H.M., Jordan, W.R., Sinclair, T.R. (Eds.), Limitations to Efficient
Water Use in Crop Production. American Society of Agronomy, Madison, WI, pp.
1–27.
Thomson, B.D., Siddique, K.H.M., Barr, M.D., Wilson, J.M., 1997. Grain legume species
in low rainfall Mediterranean-type environments. I. Phenology and seed yield.
Field Crops Res. 54, 173–187.
Timko, M.P., Ehlers, J.D., Roberts, P.A., 2007. Cowpea. In: Kole, C. (Ed.), Pulses, Sugar
and Tuber Crops, Genome Mapping and Molecular Breeding in Plants, vol. 3.
Springer-Verlag, Berlin Heidelberg, pp. 49–67.
Turner, N.C., Wright, G.C., Siddique, K.H.M., 2001. Adaptation of grain legumes (pulses)
to water limited environments. Adv. Agron. 14, 193–231.
Uppala, S., Kallberg, P., Simmons, A., Andrae, U., Da Costa Bechtold, V., Fiorino, M.,
et al., 2005. The ERA-40 re-analysis. Q. J. R. Meteorol. Soc. 131, 2961–3012.
Vadez, V., Soltani, A., Sinclair, T.R., 2013. Crop simulation analysis of phenological
adaptation of chickpea to different latitudes of India. Field Crops Res. 146, 1–9.
Whitehead, S.J., Summerfield, R.J., Muehlbauer, F.J., Coyne, C.J., Ellis, R.H., Wheeler,
T.R., 2000. Crop improvement and the accumulation and partitioning of biomass
and nitrogee in lentil. Crop. Sci. 40, 110–112.